6700k OC'd to 4.5 GHz, 40º idle 87º load with AIO NZXT x 62 kraken cooler

[iKrloz]

As the title states, i think im getting bad temps for 4.5 ghz overclock, at idle i think is cool but why if I am using a high performance liquid cooler a NZXT Kraken x62 (280 mm radiator) I am getting such bad temps a load, I cant enjoy playing Watchdogs 2 (with all settings cranked up at ultra with my GTX 1080) without getting notifications of my CPU getting to 87º in less than a minute of gameplay. I set the voltage to Adaptive since I don't really know which voltage should I set since its the first time I overclock. Also I have both turbo boost and hyper threading enabled

The things I think I should try to solve this issue is that I should set the cooler to performance mode or maybe reapply thermal paste because Im using the pre applied thermal paste that the pump had on it when I installed it, and now that I remember I changed the orientation of the pump a couple of times in order to do better cable management for aesthetics so maybe that affected the distribution of the used thermal paste.

Anyway, I would like to have your opinion before doing anything stupid, thank you in advance
My specs:
Intel 6700k OC'd 4.5 ghz with adaptive Voltage
Cpu cooler NZXT kraken x62 280mm rad liquid cooler
Mobo Asus Z170i Pro Gaming
Nvidia GTX 1080 with hybrid cooler
Corsair AX860i Power supply

 

[burnhamjs]

what voltage for Vcore is reported in the software you are monitoring with? Could be a combination of thermal paste distribution and a high level of Vcore. Also, are you running any memory overclock? (XMP Profile enabled).

 

[iKrloz]

what voltage for Vcore is reported in the software you are monitoring with? Could be a combination of thermal paste distribution and a high level of Vcore. Also, are you running any memory overclock? (XMP Profile enabled).

Yes, I forgot that detail, Im using Corsair Vengeance LED DDR4 running at 3000MHZ (speed out of the box) with XMP enabled.
Also Im not sure what the voltage is since I read that Adaptive Voltage is the most popular choice since you're not over juicing the CPU which can damage it, so the voltage will vary depending on the load. I use CPU-Z for monitoring but since Im an overclocking begginer Im nost sure what is the limit of a "safe" voltage. I'd aprecciate the feedback on that too

 

[burnhamjs]

I like to limit my Vcore to 1.35V but some are comfortable with 1.4. My 6600K is stable at 4.6G with a Vcore of 1.29V.

What max Vcore is reported in CPU-Z during stress testing? It is possible adaptive voltage is providing too much vcore and raising your temps. I recommend using a manual Vcore to see what voltage you need to be stable at 4.5G. Set it at 1.35V and see if you are stable and your temps are <80 during stress testing. If you are stable - back your voltage down to 1.34V and see if you are stable. Keep lowering until you find the minimum voltage needed to be stable. Once you find the lowest voltage go back into adaptive and see what CPU-Z is reporting as a max Vcore - then go into BIOS and apply an offset so that you are not applying more vcore then required.

Also - when I run with XMP enabled it added about 7C to my temps. On my gigabyte board there is a setting to "Relax OC" on my memory which brought my temps back down.

So - potential contributors to your high temps - bad application of the thermal interface material, vcore too high in adaptive, XMP enabled. I would start with trying to figure out the lowest vcore to be stable by using the manual vcore.

 

[iKrloz]

I like to limit my Vcore to 1.35V but some are comfortable with 1.4. My 6600K is stable at 4.6G with a Vcore of 1.29V.

What max Vcore is reported in CPU-Z during stress testing? It is possible adaptive voltage is providing too much vcore and raising your temps. I recommend using a manual Vcore to see what voltage you need to be stable at 4.5G. Set it at 1.35V and see if you are stable and your temps are <80 during stress testing. If you are stable - back your voltage down to 1.34V and see if you are stable. Keep lowering until you find the minimum voltage needed to be stable. Once you find the lowest voltage go back into adaptive and see what CPU-Z is reporting as a max Vcore - then go into BIOS and apply an offset so that you are not applying more vcore then required.

Also - when I run with XMP enabled it added about 7C to my temps. On my gigabyte board there is a setting to "Relax OC" on my memory which brought my temps back down.

So - potential contributors to your high temps - bad application of the thermal interface material, vcore too high in adaptive, XMP enabled. I would start with trying to figure out the lowest vcore to be stable by using the manual vcore.

Perfect! thank you thats an awesome and detailed reply for a beginner like me, I'll do all those tweaks when I get home And I'll keep you updated

 

[burnhamjs]

Perfect! thank you thats an awesome and detailed reply for a beginner like me, I'll do all those tweaks when I get home And I'll keep you updated

Below is something I quickly wrote up yesterday that summarizes how I overclock. There are different ways of getting to the same solution – but this is the method that works well for me. You will want to keep a particular eye on your temps considering that is currently an issue for you.

-------
Load CPU-Z so you can make sure your BIOS settings are applied and to monitor Vcore.
Load HWMonitor to keep an eye on core temps.
Load OCCT to conduct stability and max temp tests.

I would turn off turbo, EIST, and all energy saving options (c-states) while overclocking.

Set Vcore to 1.35V

Now set your multiplier to 42 and run OCCT-Small Blocks for 30 minutes and check temps. If the temps are okay (<80C), raise your multiplier to 43 and re-run. Repeat until you see your temps are exceeding 80C (or you blue screen or get an error in OCCT) - then back your multiplier down by one. While running these tests I also monitor Vcore in CPU-Z. If you see Vcore in CPU-Z is 0.01V or more under the Vcore that you set in BIOS then you are getting Vdroop and should apply some LLC. You want enough LLC to minimize Vdroop without creating Vboost. (Vboost is when you see the voltage in CPU-Z exceeding what you set in BIOS).

Now run OCCT-Large for 3 hours to make sure you can run stable. If you blue screen or get core errors in OCCT, back your multiplier down by one an re-run. Repeat until you find a stable setting for your multiplier, where you can run OCCT-Large for 3 hours.

Now that you have found a stable multiplier and temps are <80C, I like to see if I can back my Vcore down. So I lower vcore 0.01V at a time and run OCCT-Large again for 3 hours to see if I'm stable. I repeat this lowering Vcore 0.01V at a time until I get errors running OCCT-large then I put the voltage back up by 0.01V.

At this point you should find the max multiplier which you can run stable, while keeping temps <80C, with the lowest Vcore. This is your max overclock.

Now you can re-enable EIST and perhaps C-1E c-state. Also, now that you know the vcore required you could use adaptive voltage and apply any required offset to get vcore to what you need to be stable.

 

[iKrloz]

I like to limit my Vcore to 1.35V but some are comfortable with 1.4. My 6600K is stable at 4.6G with a Vcore of 1.29V.

What max Vcore is reported in CPU-Z during stress testing? It is possible adaptive voltage is providing too much vcore and raising your temps. I recommend using a manual Vcore to see what voltage you need to be stable at 4.5G. Set it at 1.35V and see if you are stable and your temps are <80 during stress testing. If you are stable - back your voltage down to 1.34V and see if you are stable. Keep lowering until you find the minimum voltage needed to be stable. Once you find the lowest voltage go back into adaptive and see what CPU-Z is reporting as a max Vcore - then go into BIOS and apply an offset so that you are not applying more vcore then required.

Also - when I run with XMP enabled it added about 7C to my temps. On my gigabyte board there is a setting to "Relax OC" on my memory which brought my temps back down.

So - potential contributors to your high temps - bad application of the thermal interface material, vcore too high in adaptive, XMP enabled. I would start with trying to figure out the lowest vcore to be stable by using the manual vcore.

Ok, so I Stressed test the CPU and the max voltage that CPU-Z was showing is 1.472, im no expert but I think that is a lot of voltage so I stopped the stress test and changed it to manual mode, at 1.32 - 1.35, Im doing the stress tests now so lets hope for the best

Now I have 2 more doubts:

1. For how long should I run the stress tests? and/or what is the recomended estimated time to do the stress tests. also, Which options should I check on the Bench tab of CPU-Z, because I saw that there are 2 boxes one for CPU single thread and other for CPU multi thread, and another one that lists some CPU references, along to two buttons: bench CPU and Stress CPU, which ones should I check and/or click?

2. When I lower the voltage to the point where gets stable at the desired 4.5 ghz, How do I limit the max adaptive voltage? is it the same as setting an offset voltage like you mentioned before? if so, how do I determine the offset voltage?

 

[burnhamjs]

Ok, so I Stressed test the CPU and the max voltage that CPU-Z was showing is 1.472, im no expert but I think that is a lot of voltage so I stopped the stress test and changed it to manual mode, at 1.32 - 1.35, Im doing the stress tests now so lets hope for the best

Now I have 2 more doubts:

1. For how long should I run the stress tests? and/or what is the recomended estimated time to do the stress tests. also, Which options should I check on the Bench tab of CPU-Z, because I saw that there are 2 boxes one for CPU single thread and other for CPU multi thread, and another one that lists some CPU references, along to two buttons: bench CPU and Stress CPU, which ones should I check and/or click?

2. When I lower the voltage to the point where gets stable at the desired 4.5 ghz, How do I limit the max adaptive voltage? is it the same as setting an offset voltage like you mentioned before? if so, how do I determine the offset voltage?

Yes, 1.472 would be too high for me. I would be more comfortable with the 1.32V-1.35. You may find stability with a lower Vcore.

1. I do not use CPU-Z for stress/stability testing. I use CPU-Z to ensure my BIOS settings were applied appropriately and to monitor Vcore. I use OCCT for stress/stability testing. I like to run OCCT-Small blocks for 30 minutes to make sure my temps are <80C. If the temps are good then I run OCCT-Large for 3 hours to ensure I’m stable. You will get varying opinions on what it takes to define “stable” as well as what program to run to check stability. OCCT is what I use.

2. I am not familiar with how adaptive voltage works in the Asus BIOS as I use a Gigabyte. I would suspect you set CPU Core Voltage to ADAPTIVE – then set you offset mode sign to “-“ then put what value you need to offset into “CPU Core Voltage Offset”.